Autonomous cars decades away, says Nissan

While the building blocks of fully autonomous vehicle technology are already available in cars on the market today, industry experts predict that we are still decades away from the widespread use of self-driving cars.

Kazuhiro Doi, Vice President, [Renault Nissan] Alliance Global Director Research and Advanced Engineering, says it will take until 2022 for the first large-scale fully-autonomous driving trials to get underway, with years of controlled research needed to validate the new systems.

"2022 is still the global plan, [and] we have not decided which countries," said at this week's Nissan Futures technology symposium in Singapore.

"Maybe we will start with normal vehicle taxi or delivery service. What we have to really think about is the use case," said Doi.

Nissan is exploring a range of Level 4 and Level 5 self-driving trials with key partners, with a view to putting fully autonomous vehicles to work in controlled environments.

Its focus is to target those locations with high need such as high-density cities as they present the best opportunity for commercialising driverless vehicles.

However, apart from the number and type of obstacles to avoid in urban environments, one of the challenges with the technology is that it is not impervious to weather conditions, so ensuring 100 per cent accuracy all of the time is something manufacturers are working towards.

"How we can secure the safety in such an environment with an unmanned vehicle is really the key," said Doi. "The weather condition is another challenge for the technologies. We use laser, LIDAR, camera, but none of the sensors can cover every condition.

"For example, a camera can't work in the dark, a radar can't work in the rain or snow, but unfortunately a customer wants to take a taxi in rain or snow conditions. The system just works on a sunny day so it cannot make a business. So that is really the challenge -- how do we expand the use case?

"Globally, Nissan has the potential to be able to introduce autonomous driving as early as 2020 but we have to be very careful."

While trials of self-driving vehicles are underway around the world, to run a full system trial with full 'eyes off' Level 5 vehicles requires cooperation between city planners, legislators, commercial operators and the public.

Just to run a 'simple' trial of self-driving taxis requires an extraordinary amount of oversight as fully autonomous vehicles require a high level of intelligence to be able to manage the unpredictable.

One such trial will commence in Yokohama, Japan on March 5, when Nissan and DeNA will conduct the first field test - including public participation - of its new jointly developed Easy Ride robo-vehicle mobility service (pictured).

Nissan's team says autonomous vehicle systems aren't yet clever enough to navigate complex pick-up and drop-off zones as a human driver would be able to do, locating a space, the correct passenger and all the permutations of a high traffic zone with multiple dangers.

Instead, self-driving trials will need to be conducted in highly controlled environments. In the case of a self-driving taxi, dedicated self-driving drop-off and pick-up zones are likely to be required.

Here, cars can more safely navigate each other and specific infrastructure would need to be in place, for example allocated collection bays so customers are identified correctly and the system can run safely and efficiently.

All of this coordination takes time, says Niels De Boer, Programme Director, Centre of Excellence For Testing and Research of Autonomous Vehicles - Nanyang Technological University.

De Boer says he doesn't see completely autonomous driving networks in widespread use for another 50 years.

"Sometimes the claims are quite bold and you have to ask: 'what do you really mean by autonomous driving?'

"You need to be realistic. You make bold statements, but you really need to ask what do you mean by autonomous driving? Because a lot of these statements are made but they don't talk about the level of maturity. Also the level of autonomy and the conditions where you drive -- you need to be very specific."

While some manufacturers have lauded their autonomous driving credentials, De Boer believes the industry should be clearer around fully-autonomous vehicle development so drivers better understand exactly what technology will be available and when.

"You read between the lines. It will be trialled during off-peak hours, it's still proof-of-concept trials. I think some of these claims are quite unhelpful and they're overly aggressive."

Singapore is developing a number of pilot locations which will be built in order to facilitate autonomous driving trials. Strict building codes will ensure roads, street signs, kerbing and lane markings are all built to specification, creating an ideal environment for unmanned vehicles to predict their environment.

Closed-circuit television and short-range communication beacons will also be installed to give researchers optimum management of the trials.

De Boer believes the most effective trials of autonomous vehicles will be those that set out to solve a particular issue. In the case of Singapore, the aim is to address the lack of parking spaces in residential areas, but even there the rules for small-scale trials will take time to get right.

"Legislation to support it is still pretty far away. If you're looking at Singapore, we're very aggressive but we have a very specific use case, and even then we're struggling with the legislation to get it ready on time."

The legal framework is also a critical enabler to the technology rollout here in Australia, where a number of autonomous vehicle trials are currently underway.

Gayle Milnes, Assistant Secretary of the Department of the Environment and Energy's Climate Change Division, said getting the landscape right for fully-autonomous vehicles requires reworking the country's driving rules and data-holding legislation.

"In Australia, work so far suggests that our legal environment is pretty good and pretty supportive of new technologies, but all of our driving laws are very focused on the driver being a human, so there's some significant changes that would be required to support these kinds of technologies going into the future," she said.

"Who owns the data, who has access, the sharing arrangements for the data, interoperability, interconnectivity between the vehicle and the infrastructure. They are among the areas of focus for us and areas that we are looking at."